Computer Model Names Agatha Christie’s Sleeping Murder as
“The Perfect Title” for a Best-Seller

Raleigh , NC (December 15, 2005) — Sleeping Murder, the last published novel by Agatha Christie, is “the perfect title” for a bestselling novel, according to an analysis of the titles of fiction bestsellers from the last 50 years.

John le Carré, however, is the novelist most consistently able to produce titles with the right attributes for success – ahead of bigger sellers like J.K. Rowling and Dan Brown.

Such are the findings of a study for Lulu.com, a website that lets anyone publish their own book.

“One of the hardest things about writing a novel is finding a good title,” says Bob Young, CEO of Lulu.com.

The study analyzed the titles of every novel to have topped the hardcover fiction New York Times bestseller list from 1955 to 2004 – and compared them with the titles of less successful novels by the same authors.

Among the study’s findings:

• Figurative or abstract titles, such as Sleeping Murder, or Presumed Innocent, produce more top-sellers than literal ones, such as The Da Vinci Code.
• A title’s length does not affect sales -- contrary to publishing industry wisdom, which decrees that bestseller titles be short.

Dr. Atai Winkler, the statistician who led the research, used about 700 titles to develop a computer model to predict the chances of a given title becoming a best-seller.

Sleeping Murder, published just after Agatha Christie’s death in 1976, was the only title to achieve the highest available score of 0.83, indicating an 83 percent probability of producing a top bestseller.

The writer with the highest career average score is John le Carré. His titles boast a 0.62 average over 19 books, of which seven have been top best-sellers.

“Our model correctly predicted whether a book was a best-seller for nearly 70 percent of the titles studied,” comments Winkler. “This is 40 percent better than random guess-work.”

“Even so,” he adds, “my advice would be to combine use of the Lulu title-scorer with using your own instincts.”

You can test the best-seller chances of your own title at www.lulu.com/titlescorer.

ABOUT LULU (www.lulu.com): Lulu, the world’s fastest growing source of print-on-demand books, enables authors and other creators to publish their own books, e-books, calendars, images, music, and films at no advance cost. It was founded by Bob Young, co-founder of Red Hat, the open source software company. Contact: pr@lulu.com.

____________________________

More detail on the Lulu study of best-selling book titles:

10 TOP-SCORING TITLES TO HAVE BEEN NYT NO. 1 BESTSELLERS

• Sleeping Murder , by Agatha Christie 0.83
• Something of Value, by Robert Ruark 0.80
• Looking for Mr Goodbar, by Judith Rossner 0.80
• Presumed Innocent, by Scott Turow 0.80
• Everything’s Eventual , by Stephen King 0.80
• Rising Sun, by Michael Crichton 0.80
• Smiley’s People, by John le Carré 0.77
• Three Fates, by Nora Roberts 0.77
• Four Blind Mice, by James Patterson 0.77
• Valhalla Rising, by Clive Cussler 0.72

10 TOP-SCORING NOVEL TITLES IN BBC’S “BIG READ” POLL

• His Dark Materials, by Philip Pullman 0.83
• Gone With the Wind, by Margaret Mitchell 0.80
• To Kill a Mockingbird, by Harper Lee 0.80
• One Hundred Years of Solitude, by Gabriel Garcia Marquez 0.77
• Ulysses, by James Joyce 0.72
• Kane and Abel, by Jeffrey Archer 0.72
• Lord of the Flies, by William Golding 0.69
• The Grapes of Wrath, by John Steinbeck 0.69
• Brave New World, by Aldous Huxley 0.69
• The Pillars of the Earth, by Ken Follett 0.69

LULU TITLE-SCORER – RESEARCH METHODS

The Lulu study was led by Dr Atai Winkler, a British statistician with both academic and commercial experience, for blue-chip clients including Tesco, Unichem and Unilever.

Winkler’s team analysed 11 different title attributes, ranging from the number of words in the title to the etymology of the words used (whether they derived mainly from Romance or Germanic languages) to whether the title contained the name of a person or place.

Of the 11 variables studied, three were found to be key ‘differentiators’ between bestsellers and non-bestsellers:

• Whether the title is literal or figurative
• The word type of the first word
• The title’s grammar pattern

Winkler’s team first segmented the data into a finite number of non-overlapping groups to find the best discriminators between bestsellers and non-bestsellers. It then applied “a probabilistic regression model” in order to find the key variables and how they should be combined into a single equation for predicting the probability of any given title producing a bestseller.